Batch Processing

Process API requests asynchronously at 50% off standard pricing. Ideal for large-scale data processing, backfill jobs, and non-real-time workloads.

Pricing

All models are available at 50% off the standard price when submitted via the batch endpoint. There is no minimum batch size. Batch jobs are typically completed within 1–6 hours, with a maximum processing time of 24 hours.

How to Use

Submit requests to the batch endpoint. Each request includes the model, messages, and any standard parameters. The system processes them asynchronously and you retrieve results via a batch ID.

# Submit a batch request
curl https://api.pandaworld.space/v1/batch \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-your-api-key" \
  -d {
    "model": "deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Hello!"}]
  }

Best Practices

• Recommended for 50+ requests to maximize throughput benefits
• Batch requests with similar prompt structures for better efficiency
• Use the batch status endpoint to monitor job progress
• Results are available for 48 hours after completion — download promptly

FAQ